diff --git a/src/graph/backends/CL/CLDeviceBackend.cpp b/src/graph/backends/CL/CLDeviceBackend.cpp
index f10eb33..92cb693 100644
--- a/src/graph/backends/CL/CLDeviceBackend.cpp
+++ b/src/graph/backends/CL/CLDeviceBackend.cpp
@@ -126,7 +126,7 @@
     ARM_COMPUTE_ERROR_ON(tensor_desc.target != Target::CL);
 
     // Create backend tensor handle
-    TensorInfo info(tensor_desc.shape, 1, tensor_desc.data_type);
+    TensorInfo info(tensor_desc.shape, 1, tensor_desc.data_type, tensor_desc.quant_info);
     auto       backend_tensor_handle = support::cpp14::make_unique<CLTensorHandle>(info);
 
     return std::move(backend_tensor_handle);
diff --git a/src/graph/backends/CL/CLFunctionsFactory.cpp b/src/graph/backends/CL/CLFunctionsFactory.cpp
index 1b448fe..ad73a79 100644
--- a/src/graph/backends/CL/CLFunctionsFactory.cpp
+++ b/src/graph/backends/CL/CLFunctionsFactory.cpp
@@ -154,10 +154,16 @@
     ARM_COMPUTE_ERROR_ON(node.num_outputs() != 1);
 
     // Extract IO and info
-    ICLTensor              *input          = get_backing_tensor(node.input(0));
-    ICLTensor              *weights        = get_backing_tensor(node.input(1));
-    ICLTensor              *biases         = get_backing_tensor(node.input(2));
-    ICLTensor              *output         = get_backing_tensor(node.output(0));
+    ICLTensor *input   = get_backing_tensor(node.input(0));
+    ICLTensor *weights = get_backing_tensor(node.input(1));
+    ICLTensor *biases  = get_backing_tensor(node.input(2));
+    ICLTensor *output  = get_backing_tensor(node.output(0));
+
+    if(is_data_type_quantized_asymmetric(input->info()->data_type()))
+    {
+        biases->info()->set_data_type(DataType::S32);
+    }
+
     const PadStrideInfo     conv_info      = node.convolution_info();
     const ConvolutionMethod conv_algorithm = node.convolution_method();
 
@@ -190,6 +196,8 @@
     // Log info
     ARM_COMPUTE_LOG_GRAPH_INFO("Instantiated " << func_name
                                << " Data Type: " << input->info()->data_type()
+                               << " Input QuantInfo: " << input->info()->quantization_info()
+                               << " Weights QuantInfo: " << weights->info()->quantization_info()
                                << " Input shape: " << input->info()->tensor_shape()
                                << " Weights shape: " << weights->info()->tensor_shape()
                                << " Output shape: " << output->info()->tensor_shape()
@@ -251,10 +259,16 @@
     ARM_COMPUTE_ERROR_ON(node.num_outputs() != 1);
 
     // Extract IO and info
-    ICLTensor                       *input         = get_backing_tensor(node.input(0));
-    ICLTensor                       *weights       = get_backing_tensor(node.input(1));
-    ICLTensor                       *biases        = get_backing_tensor(node.input(2));
-    ICLTensor                       *output        = get_backing_tensor(node.output(0));
+    ICLTensor *input   = get_backing_tensor(node.input(0));
+    ICLTensor *weights = get_backing_tensor(node.input(1));
+    ICLTensor *biases  = get_backing_tensor(node.input(2));
+    ICLTensor *output  = get_backing_tensor(node.output(0));
+
+    if(is_data_type_quantized_asymmetric(input->info()->data_type()))
+    {
+        biases->info()->set_data_type(DataType::S32);
+    }
+
     const PadStrideInfo              conv_info     = node.convolution_info();
     const DepthwiseConvolutionMethod dwc_algorithm = node.depthwise_convolution_method();
 
@@ -275,6 +289,8 @@
     // Log info
     ARM_COMPUTE_LOG_GRAPH_INFO("Instantiated " << func_name
                                << " Data Type: " << input->info()->data_type()
+                               << " Input QuantInfo: " << input->info()->quantization_info()
+                               << " Weights QuantInfo: " << weights->info()->quantization_info()
                                << " Input shape: " << input->info()->tensor_shape()
                                << " Weights shape: " << weights->info()->tensor_shape()
                                << " Output shape: " << output->info()->tensor_shape()